Wednesday, 2017-03-08

*** Sphericus has joined #openstack-powervm00:41
*** kjw3 has joined #openstack-powervm00:48
*** kjw3 has quit IRC01:05
*** esberglu has quit IRC01:07
*** esberglu has joined #openstack-powervm01:10
*** esberglu has quit IRC01:10
*** seroyer has joined #openstack-powervm01:56
*** thorst has quit IRC01:58
*** seroyer has quit IRC02:13
*** thorst has joined #openstack-powervm02:21
*** thorst has quit IRC02:23
*** thorst has joined #openstack-powervm03:24
*** thorst has quit IRC03:28
*** thorst has joined #openstack-powervm04:00
*** thorst has quit IRC04:03
*** thorst has joined #openstack-powervm05:04
*** thorst has quit IRC05:19
*** edmondsw has joined #openstack-powervm05:20
*** edmondsw has quit IRC05:25
*** Sphericus has quit IRC05:38
*** thorst has joined #openstack-powervm06:16
*** thorst has quit IRC06:20
*** thorst has joined #openstack-powervm07:17
*** thorst has quit IRC07:21
*** thorst has joined #openstack-powervm08:18
*** thorst has quit IRC08:22
*** k0da has joined #openstack-powervm08:31
*** edmondsw has joined #openstack-powervm08:57
*** edmondsw has quit IRC09:01
*** openstackgerrit has quit IRC09:03
*** thorst has joined #openstack-powervm09:18
*** thorst has quit IRC09:23
*** thorst has joined #openstack-powervm09:40
*** thorst has quit IRC09:44
*** thorst has joined #openstack-powervm10:40
*** thorst has quit IRC10:45
*** thorst has joined #openstack-powervm11:41
*** thorst has quit IRC11:46
*** smatzek_ has joined #openstack-powervm11:55
*** seroyer has joined #openstack-powervm12:26
*** thorst has joined #openstack-powervm12:33
*** seroyer has quit IRC12:46
*** seroyer has joined #openstack-powervm12:47
*** seroyer has quit IRC12:48
*** jpasqualetto has joined #openstack-powervm12:55
*** seroyer has joined #openstack-powervm13:07
*** efried has quit IRC13:08
*** edmondsw has joined #openstack-powervm13:14
*** kylek3h has quit IRC13:15
*** efried has joined #openstack-powervm13:18
*** dwayne has quit IRC13:26
*** jpasqualetto has quit IRC13:29
*** jpasqualetto has joined #openstack-powervm13:46
*** apearson has joined #openstack-powervm13:59
*** jpasqualetto has quit IRC13:59
*** kylek3h has joined #openstack-powervm14:01
*** tblakes has joined #openstack-powervm14:01
*** esberglu has joined #openstack-powervm14:07
thorsttfw you quadruple your storage capacity  :-)14:08
*** mdrabe has joined #openstack-powervm14:09
*** tjakobs has joined #openstack-powervm14:10
*** jpasqualetto has joined #openstack-powervm14:12
*** seroyer has quit IRC14:16
*** tblakes has quit IRC14:32
thorstesberglu efried: did we figure out anything with the flak8 and CI?  In openstack-nova, it looks like it also impacted the intel CI14:33
*** smatzek_ has quit IRC14:36
*** mdrabe has quit IRC14:38
*** mdrabe has joined #openstack-powervm14:38
*** seroyer has joined #openstack-powervm14:44
esbergluthorst: Not that I know of. I'm still trying to get newton/ocata CI working14:47
*** tblakes has joined #openstack-powervm14:59
*** smatzek_ has joined #openstack-powervm14:59
*** tjakobs has quit IRC15:11
*** dwayne has joined #openstack-powervm15:22
thorstwhat's up with those/15:23
esberglusetuptools keeps breaking stuff15:29
esbergluIt will be fixed once those 2 get in15:30
thorstugh, are you patching into the CI then to overcome?15:30
*** tjakobs has joined #openstack-powervm15:30
thorstefried: any reviews you want me to prioritize today?15:33
esbergluI haven't yet. I can if you want, but it will require an image rebuild, then probably another image rebuild after the changes merge15:33
thorstis there another option?  Or if you think they'll merge soon, we could just wait it out.15:34
esbergluI'm hoping they merge soon. Pretty trivial changes, just excluding certain setuptools versions15:40
esbergluThere's currently only 1 change out on those branches15:42
esbergluFor the *-powervm projects15:42
*** mdrabe has quit IRC15:43
esbergluwhich is the only thing waiting for the merge will block15:43
*** tjakobs has quit IRC15:51
efriedthorst Sorry, not sure why notifications aren't popping up all the time here.  I figured out the flake8 thing.  Will try to get it resolved today.15:52
efried is what broke us.15:52
*** mdrabe has joined #openstack-powervm15:52
efriedthorst Review-wise, it would be good if you could take a first pass at
efriedI might want to spend some time trying to figure out how to reduce the LOC count there.15:53
efriedI already pared a bunch out by ripping out the base disk adapter.15:53
efriedAnd I still need to add more UT.15:53
thorstk.  I'm looking over one from taylor quick then I'll get to that15:58
*** gaperez has joined #openstack-powervm16:14
*** gaperez has quit IRC16:14
thorstefried: damn...that is a large patch16:16
*** jpasqualetto has quit IRC16:17
*** k0da has quit IRC16:31
thorstefried adreznec esberglu: Should we W+1 these?
thorstis there a reason not to?16:32
*** jpasqualetto has joined #openstack-powervm16:34
adreznecthorst: Not that I'm aware of16:35
thorstefried did +216:35
thorstso I guess just waiting for esberglu to say it won't make his life awful16:35
esbergluWhy would it make my life awful?16:36
thorstdon't think it would16:36
thorstbut you're battling CI, a world where anything can go wrong  :-)16:36
esberglucan and eventually will16:37
thorstadreznec: do we know if the pypowervm version stuff is being worked still?16:37
adreznecthorst: Was going to ping Julio about that later today16:38
*** tjakobs has joined #openstack-powervm16:39
*** openstackgerrit has joined #openstack-powervm16:42
openstackgerritMerged openstack/networking-powervm master: Bump to match latest global requirements
openstackgerritMerged openstack/ceilometer-powervm master: Bump to match latest global requirements
efriedesberglu Does the CI env have nvls with dual VIOS where each VIOS is on a separate SSP?16:55
esbergluNo everything CI is single vios16:55
efriedthorst - The base disk adapter is already not in the SSP change set.16:58
efriedSo I'm confused about your comment.16:59
thorstefried: I'm saying make two change with it16:59
thorstcan we get half the code into the base disk adapter16:59
thorstor no?16:59
efriedOh, I gotcha.  The first change wouldn't do anything.16:59
efriedBut... not really any savings by doing that.17:00
thorstwell, so I think the core team was OK if we said 'here's the disk patch series'17:00
thorstafter the third patch, it'll work.17:00
efriedYeah, I understand.17:00
thorstso was trying to figure out a sensible way to decompose into that...17:00
efriedYeah.  I'll take a look at it.  Try to get it under 500LOC per change set.17:00
efriedThough there's really not an awful lot of code in here.  A bunch of docstrings/comments.  A bunch of UT.17:01
*** jpasqualetto has quit IRC17:03
thorstefried: yep...17:07
*** efried has quit IRC17:13
*** jpasqualetto has joined #openstack-powervm17:16
*** efried has joined #openstack-powervm17:21
efriedthorst So I'm gonna rebase all the change sets.17:25
efriedI suspect I'm going to need to do that in order to get this flake8 thing resolved.17:25
efriedthorst adreznec - Any reason I need to bump the pypowervm req in nova/requirements.txt?17:29
efriedNo functional difference between .4 and .4.1; leaving it as .4 would work with the new g-r because it's >=17:30
adreznecefried: no, I don't think17:32
adreznecthey should eventually get the new req from a g-r sync17:32
openstackgerritDrew Thorstensen (thorst) proposed openstack/nova-powervm master: Switch to qcow handler
thorstefried adreznec seroyer: ^^17:35
*** kjw3 has joined #openstack-powervm17:36
efriedthorst Hey, btw, mriedem gave me a bit of grief yesterday for our use of TaskFlow.17:43
efriedI had been under the impression that it was The OpenStack Way.17:43
efriedBut when he expressed surprise that we were using it, I went and checked.  There's nothing else in the nova project that uses it.17:43
efriedSo - where did that come from?17:44
thorstfrom OpenStack was my understanding.  There was a desire from various projects around the Atlanta summit (way way back) to use task flow for things...especially when things go wrong17:48
thorstthere was a push to use it, but the issue was always "we have so much code, how to go back and reimplment"17:48
thorstand also, when you just have a bug to fix, you don't want to go back and resolve all the things17:49
thorstbut since we were writing code from scratch...and we've done this rodeo before...I wanted to use it17:49
thorstand I am so freaking thankful we did.  Yeah, it adds a layer, but when sh!t goes gets cleaned up right.17:49
thorstwe had a lot of arguments way back then about the complexity it added...which is legit...but its really reduced the defects.17:51
*** apearson has quit IRC17:56
*** smatzek_ has quit IRC17:59
*** apearson has joined #openstack-powervm18:00
efriedthorst May be worth having that conversation with mriedem.18:11
efriedWas he around at the Atlanta summit?18:11
efriedOr was he, like, twelve at the time?18:11
*** smatzek_ has joined #openstack-powervm18:20
thorstlol, I don't know if he was there18:20
thorstI'll chat with him18:21
thorstotp with a user though, so I'll ping him when I get off the call18:22
openstackgerritTaylor Jakobson proposed openstack/nova-powervm master: Add support for file-backed ephemeral disk
*** apearson has quit IRC18:32
*** apearson has joined #openstack-powervm18:39
thorstefried: if you get a chance...I'd like to push this through today:
thorstremoves the need for a patch that the PowerVC team is carrying.19:36
*** k0da has joined #openstack-powervm19:42
*** dwayne has quit IRC20:02
*** apearson has quit IRC20:26
*** apearson has joined #openstack-powervm20:28
efriedthorst +2 - but I thought the REST layer was going to switch to this by default?20:47
thorstno, REST was not going to change20:47
thorstopenstack was to change20:47
adreznecI was actually just looking at that patch thorst20:48
adreznecAre we intending to basically only support the qcow handler for now and patch in a configopt for other handlers in the future?20:49
thorstadreznec: I'm not sold that we need other handlers in the future...I was operating under assumption that openstack would really only need one type20:51
thorstthough, I could be sold that maybe the cinder connector (ex. or should choose which handler to use?20:51
adreznecthorst: is the hope that we could autodetect based on filetype or something?20:51
thorstnot really, I guess I just don't want to have to figure it out per type...just seems like excess configuration.20:52
thorstdo we need multiple types?20:52
adreznecI mean that works for things like qcow vs raw, but makes things like ceph ephemeral fuzzy20:52
adreznecDepends on where we go I suppose20:52
thorstso this is the qcow handler, not the qcow file type...right?20:52
adreznecNot for just qcow vs raw20:52
adreznecBut down the road if we want to support other ephemeral types20:52
thorstso maybe we cross that bridge when we get to it?  Other connectors needing different types?20:53
thorstwell, and its not just change is for cinder20:53
thorsttjakobs is for nova ephemeral20:53
adreznecRight sorry, was looking at his patch now and mixed up wording :P20:53
thorstso I don't think its fair to ask the administrator what 'handler' do you want to use20:54
adreznecI think that's fair20:54
thorstI think its fair to say 'are you using gpfs, nfs, or ceph (or whatevs)'20:54
thorstand then pick the handler based off that20:54
thorsteven for tjakobs change...I think maybe we need an enumeration there...even though right now we're just bundling it all as 'fileio'20:55
thorstadreznec: if you look through the cinder volume drivers you see any that shouldn't be qcow2?20:55
adreznecthorst: not for file-backed volumes20:57
adreznecI mean other volume types like iSCSI won't be of course, but they're just block20:57
*** dwayne has joined #openstack-powervm20:57
thorstright...that's completely different21:00
thorstso I think for tjakobs change, I don't think I want 'fileio' to show up as a disk type21:00
openstackgerritMerged openstack/nova-powervm master: Switch to qcow handler
thorstI think it should be local_file...21:01
thorstmake the option something that an admin can understand.  Because fileio indicates 'anything that is a file'...which isn't necessarily true?21:01
thorstwhat use cases are we targeting for his change set?  Local files, NFS?21:02
tjakobsso you want the nova.conf option to be "local_file" instead of "fileio"?21:06
tjakobsI've also only tested the patch on local files21:06
adreznecthorst: do we actually care what the backing storage the files sit on is in that case?21:08
adreznecI mean putting in on some parts would certainly have a performance penalty21:08
thorstadreznec: I thought you said it would/could?  :-)21:08
adreznecBut if Linux has it mounted as a filesystem you can put stuff on...21:09
thorstso, maybe we name it something to just indicate a standard filesystem21:09
adreznecIt would if we're using a specialized handler21:09
thorstmaybe just 'filesystem'21:09
adreznecLike Ceph21:09
adreznecThat has tighter coupling with the backing devices21:09
adreznecBut if we're just slapping a file down and mapping it21:10
* adreznec shrugs21:10
adreznecShould all look the same21:10
thorstso I guess instead of fileio...we go with filesystem?21:11
thorstor just file21:11
thorsttjakobs: your call on which.21:11
tjakobsok, i'll toss that in the next update I do21:12
efriedtjakobs thorst apearson adreznec - Does the FileIO object need to be PUT to the REST server?21:15
thorstthe file itself? no.  The path to the file21:15
efriedSo I'm looking at tjakobs change set, and seeing that we're doing all of these ops directly on the file system.  I would have thought, for the sake of API consistency, that the REST API should be doing those ops instead, and the community code should be going through that.21:16, I'm not so sure21:18
thorstwe've done the whole stream a file from NL to REST before21:18
thorstso there is precedent having things on NL.21:18
thorstcfg drive for instance?21:18
thorststreaming through rest would be a lot of overhead, and I'm not sold its consistent21:19
adreznecyeah efried I was just looking at his changeset too21:24
*** apearson has quit IRC21:24
efriedthorst Even when we do the image upload "directly", it's to a path retrieved from the REST API.21:25
thorstyeah, but I think the whole idea of fileio isn't that we stream the file through rest...just that we point rest to it.21:26
thorstwe probably should confer with apearson, but I think that's the design intention.21:26
adreznecCalling things like qemu-img directly(ish) through the nova utils is again breaking that model21:26
efriedRealistically, we would still be composing that path from CONF.instances_path etc. - but doing that once and then asking the REST API for that path subsequently for ops like the upload.21:26
efriedI'm more concerned with stuff like creating, deleting, and resizing the file on the file system.21:26
thorstI guess I'd have the concern if we were talking volumes...but there was a line drawn between volumes and files to me.21:28
thorstwhere basically you just tell REST the file...and you do whatevs to it (ex. cinder driver can clone it)21:28
thorstI need to relocate to house a little early today21:30
thorstback in a bit21:30
*** thorst has quit IRC21:30
*** thorst has joined #openstack-powervm21:31
*** smatzek_ has quit IRC21:31
*** thorst has quit IRC21:35
*** apearson has joined #openstack-powervm21:42
*** thorst has joined #openstack-powervm21:50
*** kjw3 has quit IRC22:20
*** edmondsw has quit IRC22:20
*** edmondsw has joined #openstack-powervm22:23
*** edmondsw has quit IRC22:28
efriedesberglu Seeing quite a lot more in-tree CI failures recently.22:29
efriedDo you know what that's about?22:29
esbergluLooking at it right now22:29
esbergluRemember those rebuild server tests that were hitting us before?22:29
esbergluThey seemed to stop for a while, but they are hitting us again now22:30
efriedActually right now this is the only one I care about that's gotten a failure:
esbergluThere were 3 rebuild tests that were failing. We disabled the rebuild from error because it seemed to just be taking too long22:32
esbergluAnd then I didn't see the other two (the ones hitting us now) for a while, so I thought they were a side effect from the 1st one22:33
efriedthorst adreznec esberglu - Need re-+1s on rebased-and-slightly-updated change sets:
*** dwayne has quit IRC22:44
*** tblakes has quit IRC22:47
*** esberglu has quit IRC22:51
*** esberglu has joined #openstack-powervm22:52
*** thorst has quit IRC22:54
*** thorst has joined #openstack-powervm22:55
*** esberglu has quit IRC22:56
*** mdrabe has quit IRC22:56
*** edmondsw has joined #openstack-powervm22:57
*** thorst has quit IRC22:59
*** edmondsw has quit IRC23:01
*** tjakobs has quit IRC23:09
*** kylek3h has quit IRC23:11
*** tblakes has joined #openstack-powervm23:12
*** thorst has joined #openstack-powervm23:20
*** tblakes has quit IRC23:24
*** thorst has quit IRC23:24
*** jpasqualetto has quit IRC23:25
*** seroyer has quit IRC23:35
*** jpasqualetto has joined #openstack-powervm23:41
*** jpasqualetto has quit IRC23:42
*** apearson has quit IRC23:45

Generated by 2.14.0 by Marius Gedminas - find it at!